Downloaded from http://injuryprevention.bmj.com/ on June 18, 2017 - Published by group.bmj.com
252
Injury Prevention 1997; 3: 252-256
ORIGINAL ARTICLES
Home drowning
children
among
preschool
age
Mexican
Alfredo Celis
ground water exist, these are insufficient to
satisfy the needs of the population. The most
important source is the Lake of Chapala, about
50 kilometres away, from which water is
pumped. However, cuts are frequent, particularly in the 'dry' season (approximately four
months every year), during which the rain
diminishes and the natural deposits of water
diminish. Because of this many people store
water in wells, barrels, tubs, pails, and underground cisterns ('aljibes'; see figure). Due to
economic conditions, in this region houses
with swimming pools or bathtubs are scarce.
Several studies show that almost any body of
water at home poses a risk for drowning. Those
most frequently reported are swimming pools,
bathtubs, and buckets.8 13-18 However, most of
these studies come from developed countries
whose economic conditions are very different
from Mexico. Particularly in the Metropolitan
Area of Guadalajara (the second main urban
concentration in Mexico) it has been observed
that approximately 60% of drowning in children aged 1 to 4 years occurred in underground cisterns or aljibes.'2 Other bodies of
water associated with drowning in this age
group include water wells or draw wells
(17.6%), tubs or buckets (11.3%), swimming
pools (4.4%), and basin fronts (courtyard
pools to store water) (4.4%) (A Celis, unpublished data). None of these has been evaluated
as a risk factor in these localities and some have
never been reported in the international
literature.
(Injury Prevention 1997; 3: 252-256)
Accordingly, the aim of this study was to
estimate the relative risk for drowning in the
Keywords: drowning; home injuries; Mexico; risk 1 to 4 year age group, presented by different
factors.
bodies of water found in or near Mexican
homes.
Abstract
Objectives-To estimate the risk of
drowning by different bodies of water in
and near the home for children aged 1 to 4
years.
Setting-The Metropolitan Area of Guadalajara, Mexico.
Methods-A population case-control
study. Cases (n=33) were children 1 to 4
years old who drowned at their home;
controls (n=200) were a random sample of
the general population.
Results-The risk of drowning for children
whose parents reported having a water
well at home was almost seven times that
of children in homes without a water well
(adjusted odds ratio (OR)=6.8, 95% confidence interval (CI)=2.2 to 20.5). Risk
ratio estimates for other bodies of water
were: swimming pools (OR=5.8, 95%
CI=0.9 to 37.5), water barrel (OR=2.4,
95% CI=1.0 to 5.6), underground cistern
(OR=2.1, 95% CI=0.8 to 5.2), and a basin
front (courtyard pool to store water) of 35
or more litres (OR=1.8, 95% CI=0.8 to
4.4).
Conclusion-Drowning at home is frequent in the Metropolitan Area of Guadalajara, but the causes are different from
those reported in developed countries.
Accordingly, the preventive strategies
must also be different.
Sierra Mojado #950,
Colonia Inderendecia
Centro Medico,
44340, Guadalajara,
Jalisco, Mexico
Correspondence to:
Dr Celis.
Due to its high incidence in many places
around the world, drowning is an important
public health issue. It is often most frequent in
the 1 to 4 year old group'-6 where annual rates
are from 5/100 000 to 13/100 000.78 As with
many other kinds of injury, males are at higher
risk.9 10 In Mexico, from 1985 to 1990, 17 092
deaths by injury occurred among children aged
1 to 4 years. Of those 3408 events (19.9%)
were drownings (International Classification of
Diseases E9 10), placing it first in the mortality
list."' As well as the high frequency of these
events, another important aspect is that 63.1 %
happen at home.'2
The Metropolitan Area of Guadalajara
includes approximately three million inhabitants. This population is located 1500 meters
above sea level. Although deposits of under-
Methods
A case-control study was conducted. Cases
were children 1 to 4 years of age who were
inhabitants of the counties that compose the
Metropolitan Area of Guadalajara (Guadalajara, Zapopan, Tonala, and Tlaquepaque) and
who were fatally drowned at their home
between 1991 and 1993. All cases were
identified from death certificates provided by
the statistics office of the Jalisco State Public
Health Department. Additional information
was obtained by interview with the mother.
The date of death was taken as the reference
date, and all other information was related to
this date.
Controls were children randomly selected
from among the urban residents living in the
253
Home drowning among preschool age Mexican children
..:::
1.11"Mr.,
...
.:;
An aljibe
or
underground cistern.
Metropolitan Area of Guadalajara, matched
by sex and age at the time of the reference
date of the case (for example, if the case was
2 years old on the date of the drowning, the
control had to be between his or her 2nd and
3rd birthday). For this random selection we
used the Metropolitan Area of Guadalajara
land register charts by means of the following
algorithm in which the house selection was
made in two stages: (a) selection of the block
and (b) selection of the home. By using the
codes contained in the land register, the
block was chosen with the aid of randomly
selected numbers that pinpoint the land
register area, the block, and the sidewalk.
From there, all houses in that side of the
block were numbered and one was selected
randomly.
At each selected home we proceeded as
follows: after the presentation and before the
initiation of the questionnaire, the interviewer
asked if there was another child of the same sex
and age living there at the time of death of the
case. If not, the next home to the right was
chosen, or, in the case of an apartment
building, the next greater number was used,
until an eligible control was identified.
Interviews were conducted by eight social
workers trained in the use of a standardized
questionnaire. In all but two interviews the
data were obtained directly from the mother of
the study child. For the cases, all questions
were posed with reference to the date the
drowning occurred. For the controls, the
questions were made using the reference date,
as above.
The variables investigated included characteristics of the child (sex and age); the mother
(age, marital status, occupation, education);
the head of the household (age, sex, occupation, education); the family (monthly family
income, number of members of the family); the
household (walls, property, type, construction); and bodies of water (underground
cistern, bathtub, buckets and/or tubs, barrels
of water, basin front with 35 or more litres,
water well, swimming pool). Families of the
cases were also asked to describe how the
drowning occurred.
Three bodies of water need explanation.
Underground cisterns are shaped like a room,
with walls, roof, and floor made of concrete.
They have a small opening (about 50 x 50 cm)
at the top, which is at the same level as the floor
of the house. Their capacity varies from 500 to
1000 litres. By barrel we mean a container
made of steel, about 200 litres in capacity, into
which many liquid substances are delivered to
industries (oil, gasoline, etc). In poor areas
they are popular. The cover is removed and
water is stored in them. Basin fronts (previously explained) are found in every house,
but the largest are found in old houses. The
water capacity varies, but can reach 200 litres
or more. Basin fronts are usually used to wash
clothes at home.
Cases were compared with controls to
estimate the relative risk of drowning by each
body of water. Matched and unmatched
analyses were performed and the results
compared. Because they were similar, only
unmatched results are presented. The odds
ratio (OR) and 95% confidence intervals (95%
CI) were computed for each risk factor using
logistic regression'9 except for two variables
that had three categories of outcome. For
these, the Mantel-Haenzsel procedure was
used and the 95% CI was calculated by the
exact method.
Stratified analyses were used to control for
confounding by covariates and to assess interactions. The confounding effects of education
(schooling of the head of the household) and
brick as the main wall building material
(associated with socioeconomic status) were
both adjusted. After these were in the model,
no further adjustment was necessary; nevertheless, the mother's education was included
when examining the effect of the education of
the head of the household as a risk factor for
drowning. Goodness of fit was assessed graphically by delta-beta values, which are an
approximation of the amount that an estimated
regression parameter would change if a given
observation were omitted from the regression
fit.20
Results
From January 1991 to December 1993, 65
home drownings were identified. Data were
obtained from 33 cases: of the remainder nine
were excluded because the event did not occur
Table 1 Bodies of water at home in which the child fell
Bodies of water
Frequency (%/6)
13 (39.4)
Aljibe (underground cistern)
Bucket/tub
Water well
Barrel
Pot
Bathtub
Puddle
Swimming pool
Basin front
7
5
2
2
1
1
1
1
(21.2)
(15.2)
(6.1)
(6.1)
(3.0)
(3.0)
(3.0)
(3.0)
Downloaded from http://injuryprevention.bmj.com/ on June 18, 2017 - Published by group.bmj.com
254
Celis
in the home; nine could not be located because
the family moved; one could not be followed
up after three home visits; and 13 were not
located due to wrong addresses. Thus, ap-
Table 2 Characteristics of the mother and head of the family: cases and controls
Characteristics of the mother
Age (years)
<20
>20
Marital status
Unmarried
Married
Employment
Notworking
Working
Schooling
Elementary incomplete
Elementary completed
OR
95%
OR
CI*
adjustedt
95%
CI*
3.6
1.0
1.0 to 12.6
1.2
0.2 to 6.3
14
184
2.5
1.0
0.8 to 7.6
1.2
0.3 to 4.3
23
8
142
58
1.2
1.0
0.5 to 2.8
1.4
0.5 to 4.0
14
17
43
157
3.0
1.0
1.4 to 6.6
1.4
0.5 to 3.8
2.2
0.8 to 6.4
5.1$
2.0to 1.32
Cases
Controls
cnrde
4
27
8
192
5
26
Characteristics of the head of the family
Age (years)
<25
8
26
2.3
0.9 to 5.7
>25
174
23
1.0
Schooling
Elementaryincomplete
16
33
5.4 2.4to 11.9
Elementary completed
15
166
1.0
*CI by el±1(9e(fs))
tAdjusted for schooling of the head of the family and walls of the house.
$Adjusted for schooling of mother and the walls of the house.
Table 3 Characteristics of the family and home cases and controls
Characteristics of the family
Monthly family
income (US$)
<315
315-524
525
No of inhabitants
<- 3
4-6
->- 7
Cases
Controls
OR
crude
95%
CI*
OR
adjustedt
95%
CI*
13
12
7
66
58
70
2.0
2.1
1.0
0.7 to 5.9t
0.7 to 6.3t
1.2
1.5
1.0
0.3 to 4.8t
0.4 to 5.7$
4
19
9
42
112
46
1.0
1.8
2.0
0.5 to 7.6$
0.5 to 9.8t
1.0
1.9
0.5
0.5 to 11.0t
0.1 to 4.5t
0.1§
0.02 to 0.3
1.2
0.5 to 3.2
1.2
0.5 to 3.0
Characteristics of the home
Materials of the walls
Brick
24
192
0.1
0.03 to 0.3
Other
7
5
1.0
Home property
Owned
18
127
0.7
0.3 to 1.6
Rented
14
73
1.0
Home under construction
Yes
12
44
1.9
0.9 to 4.2
No
20
141
1.0
*CI by efl 96(8e(O))
tAdjusted for schooling of the head of the family and walls of the house.
tCI by exact calculation.
§Adjusted for schooling of the head of the family.
Table 4 Characteristics of the bodies of water: cases and controls
OR
95%
Cases
Controls
crude
CI*
OR
adjustedt
95%
CI*
2.1
0.8 to 5.2
0.7
0.2 to 2.9
0.6
0.2 to 1.5
2.4
1.0 to 5.6
1.8
0.8 to 4.4
6.8
2.2 to 20.5
5.8
0.9 to 37.5
Aljibe (underground cistern)
Yes
20
107
1.3
0.6 to 2.8
No
13
93
1.0
Bathtub
Yes
4
25
1.0
0.3 to 2.9
No
29
173
1.0
Bucket/tub 20 or more litres
Yes
22
148
0.7
0.3 to 1.5
No
11
52
Barrel of water
Yes
16
53
2.6
1.2 to 5.5
No
17
147
1.0
Basin front 35 or more litres
Yes
16
60
2.2
1.0 to 4.6
No
17
140
1.0
Water well
Yes
9
12
5.9 2.2 to 15.4
No
24
188
1.0
Swimming pool
Yes
2
4
3.1 0.6 to 17.9
No
31
195
1.0
*CI by el±s.96(se(IS))
tAdjusted for schooling of the head of the family and walls of the house.
proximately 60% of the cases eligible for the
study provided data for the analysis. Using the
neighbourhood as an indicator of socioeconomic status, it appears socioeconomic status
did not differ between missing and study cases.
All cases were compared with 200 controls.
In general, the proportion of non-response was
less than 10%, but this differed according to
social group: in neighbourhoods of high socioeconomic status the rate of non-response was
greater than in those of lower socioeconomic
status. Unfortunately, the questionnaires were
not designed to register this non-response data.
The places where drownings most often took
place are shown in table 1: they include
underground cisterns (39.4%), buckets/tubs
(21.2%), and wells (15.2%).
Crude and adjusted analysis of the mother's,
family's head, and household characteristics
for cases and controls are compared in tables 2
and 3. The mothers of cases were more likely
to be young (adolescents), to live as a couple
with someone but unmarried, without paid
employment, and were less well educated.
Similarly, the family heads of the cases were
more likely to be young and with only primary
school education. Case families reported lower
family income, more family members, and
more often lived in houses under construction
than controls.
The crude and adjusted risk factors for
drowning by different bodies of water are
presented in table 4. The presence of wells
was the strongest risk factor, with almost six
times the risk (OR 5.9; 95% CI 2.2 to 15.4) of
those who did not have a well. The second
strongest association was with swimming pools
(OR 5.8; 95% CI 0.9 to 37.5). Weaker
associations were found with barrels of water
(OR 2.4; 95% CI 1.0 to 5.6) and underground
cisterns (OR 2.1; 95% CI 0.8 to 5.2), while
bathtubs, buckets/tubs, and basin fronts did
not show any significant association.
From table 4 we can also estimate the
prevalence of different bodies of water found
in Guadalajara. As we see from the controls,
53.5% of households have underground cisterns, 12.6% have a bathtub, 74.0% a bucket/
tub of 20 or more litres, 20% a barrel of water,
30% a basin front of 35 or more litres, 6.0% a
well, and 2.0% a swimming pool. The prevalence of all bodies of water except bathtubs
and buckets/tubs were higher in case households: underground cisterns in 60.6%, bathtub
in 12.1%, buckets/tubs in 66.7%, barrel of
water in 48.5%, water well in 27.3%, and
swimming pool in 6.1%.
Discussion
This study identifies different bodies of water,
some not reported previously, in which children 1 to 4 years old drown at home. It
provides estimates of the association between
drowning and the presence of these deposits.
Several characteristics of the home revealed in
this study need to be considered when formulating preventive measures to diminish the
frequency of this public health problem.
The study shows that almost any water
Downloaded from http://injuryprevention.bmj.com/ on June 18, 2017 - Published by group.bmj.com
255
Home drowning among preschool age Mexican children
deposit represents a risk factor for drowning
among preschool age children. Due to their
frequency or strong association, several demand special attention, such as wells and
residential swimming pools. However, because
these are scarce in Guadalajara this explains
why only six children drowned in them. A less
strong association was found with barrels, large
basin fronts, and underground cisterns. The
latter are popular in Guadalajara and are where
most children drown. Because of their high
prevalence among controls, however, the association is not as strong as that for wells.
Nevertheless, the number of casualties are
large enough to warrant a study to identify
which part of the cisterns is related to these
drownings.
These data do not allow us to measure the
drowning risk that buckets and tubs represent.
Jumberlic and Chambliss suggested that buckets with a capacity of 20 or more litres
represent a major risk,"6 while our results
suggest that buckets of this size could represent
the same risk as those with a smaller capacity.
Without affecting their functional usefulness,
such buckets could be designed with an
unstable base so that when a child falls into
one, the bucket tips easily, allowing the water
to spill.
Our results also show that asphyxia by
immersion in the home is strongly associated
with the socioeconomic conditions of the
family. Although family income does not show
an important association, the educational level
of the head of the household and the characteristics of the home (that also indicate social
status) stand out as risk factors. In Guadalajara, families of very low socioeconomic status
use material other than brick (or other stonelike materials) for home construction. It is of
particular importance to highlight that this
characteristic is a proxy that may allow us to
identify groups with an increased risk of
drowning.
Several limitations require consideration.
Not being able to locate some cases influences
the results. We believe, however, that this
results in a bias toward the null. A previous
study in the same city found that 56.1% of
drownings at home happened in underground
cisterns,'2 while in this study only 39.4% of
cases reported being drowned in an aljibe. It
seems that families of cases that drowned in
underground cisterns change their address
more frequently than controls. This would
bias our results toward the null hypothesis,
because the frequency in which an aljibe is
reported for a case would approach that of
controls. We do not have previous reports for
other bodies of water, but it is reasonable to
assume that if families move because they do
not wish to live in a place where their child
drowned, all other risk estimates will also tend
to the null.
A second limitation is that the data collected
were based on the informant's memory. It
could be that mothers of cases remember more
about water systems at home than mothers of
controls. However, almost all bodies of water
can be observed, and their presence and
characteristics remain for a long time. On the
other hand, the absence of previous home
drowning studies prompt us to assume that
recall bias would be small.
Controlling for confounding could be incomplete. Before this study there were no
reports to guide the associations being examined. Possible confounding factors such as age,
sex, and indicators of social status have been
reported in relation to other injury deaths.
Cases and controls were matched by age and
sex, and the results show that these did not
have a confounding affect. Correlates of social
condition were used to control for their
obvious effects in the statistical analysis.
Finally, although it was impossible to prevent the interviewers from knowing the case or
control status or the subjects interviewed, and
this could introduce an information bias, to
diminish this we used a standardized questionnaire and trained interviewers.
The study also has several advantages that
merit attention. The control group was
randomly selected from the community, thus
enabling us to estimate the prevalence of
bodies of water at home and relative risks
for drowning that have not been reported
previously. Moreover, the cases and controls
were population based and this diminishes
the probability of selection bias. As the case
and control interviews were identical, the
risk of differential misclassification is lessened.
Implications for prevention
This investigation shows that the problem of
drowning at home by 1 to 4 year old children in
the Metropolitan Area of Guadalajara,
although occurring with the same frequency
as reported by other investigators, involves
different aetiological characteristics than those
reported previously in developed countries.
Accordingly, the preventive strategies for Guadalajara must also be different. The characteristics of bodies of water not reported
previously, and their relation with drowning,
should be studied in depth.
1 Fingerhut LA, Kleinman JC, Malloy MH, Feldman ni.
Injury fatalities among young children. Public Health Rep
1988; 103: 399-405.
2 Barell V, Zadka P, Halperin B, Sidransky E. Childhood
mortality from accidents in Israel, 1980-84. IsrJMed Sci
1990; 26: 150-7.
3 Keeling JW, Golding J, Millier HKGR. Non-natural deaths
in two health districts. Arch Dis Child 1985; 60: 525-9.
4 Knobel GJ, de Villiers JC, Parry CD, Bith JL. The causes of
non-natural deaths in children over a 15-year period in
greater Cape Town. S Aft Med J 1984; 66: 795 - 801.
5 De Mello MH. Mortes violentan em menores de 15 anios no
Brasil. Bol Cfcina Sanit Panam 1986; 100: 590 - 603.
6 Informaci6n Estadistica Sector Salud y Seguridad Social.
Mexico: Instituto Nacional de Geografia Estadistica e
Informatica, 1989.
7 Division of Injury Control, CDC. Childhood injuries in the
United States. Am J3 Dis Child 1990; 144: 627-46.
8 Pearn JH, Wong RYK, Brown III J, Ching Y, Bart R,
Hammar S. Drowning and near-drowning involving
children: a five-year total population study from the city
and county of Honolulu. Am J Public Health 1979; 69:
450-4.
9 Dietz PE, Baker SP. Drowning. Epidemiology and prevention. Am J Public Health 1974; 64: 303 - 12.
10 Mizuta R, Fujita H, Osamura T, Kidowaki T, Kiyosawa N.
Childhood drownings and near-drownings in Japan. Acta
PaediatrJ7pn 1993; 35: 186-92.
1 1 Bases de datos de certficados de defunci6n. Mexico 1985- 1990.
INEGI-Secretaria de Salud.
Downloaded from http://injuryprevention.bmj.com/ on June 18, 2017 - Published by group.bmj.com
256
Celis
12 Celis A. Asfixia por inmersion en Jalisco, 1983 - 1989. Salud
Publica Mex 1991; 33: 585-9.
13 Budnick LD, Ross DA. Bathtub-related drowning in the
United States, 1979-81. Am J Public Health 1985; 75:
630-3.
14 O'Carroll PW, Alkon E, Weiss B. Drowning mortality in
Los Angeles County, 1976 to 1984. JAAMA 1988; 260:
380-3.
15 Peam J, Nixon J, Wilkey I. Freshwater drowning and neardrowning accidents involving children. A five-year total
population study. Med JAust 1976; ii: 942-6.
16 Jumberlic MI, Chambliss M. Accidental toddler drowning
in 5-gallon buckets. JAMA 1990; 263: 1952-3.
17 Gardiner SD, Smeeton WM, Koelmeyer TD, Caims FJ.
Accidental drowning in Auckland children. NZ Med J
1985; 98: 579-82.
18 Mann MC, Weller SC, Rauchshwalbe R. Bucket-related
drownings in the United States, 1984 through 1990.
Pediatrics 1992; 89: 1068-71.
19 Hosmer DW, Lemeshow S. Applied logistic regression. New
York: Hohn Willey and Sons, 1989.
20 EGRET. Reference manual. Seattle, WA: SERC, 1992.
A BOY DROWNING.
Hundreds of poor boys are drowned every
year from not being sensible of the danger of
water,r They *oiltopon(ls and r iers wItlout
knowing their dlepth, and by one fatal step they
sinlk, iever to rise again. Bovs slhould never
bathc but in badts madle for the p)urp)o)se, or
s11ldd be attenled by those who canl help them
ifr ttt'lte-v.Iclt 'we have a picture or two
little l)s4x;i0o wvetit oiut on ;afternoon to bathee:
fine of thietn vtittired(tut too frtr, alnld we see
himl 4f1i the point of sinking, extendninlg his arms
tbiwards his briotlher, 1ll 11 )pes of bLing rescuet
fromoli a wvatcr -rz.tve. What sorrow rests upon liko cotitetanaiice, as hie thinks oif lovinig pareits,t, kindl -frnjemls, and1flUcctionate brothers
alnil%l
sister! Iltit we hwpe lhe will ie saved,
asil jlrove a tsefid member of sociely. Ali! a
hlpts lbandt i; neaar! 'The stranl11fo' positionl of
hi. lbr-iherl1tr atl radvtl, thle liotice ofr a ]b nevoilrtt
istransg.'r, who r ushaed intto tle water Stitil Sa;'4l
|the *l1r.IrUIB bUy, at tle 'itoltletlt risk *f Ili
'
*MVIll
l.
From The Book of Accidents; Designed for Young Children,
New Haven, 1830.
Downloaded from http://injuryprevention.bmj.com/ on June 18, 2017 - Published by group.bmj.com
Home drowning among preschool age
Mexican children.
A. Celis
Inj Prev 1997 3: 252-256
doi: 10.1136/ip.3.4.252
Updated information and services can be found at:
http://injuryprevention.bmj.com/content/3/4/252
These include:
Email alerting
service
Receive free email alerts when new articles cite this article. Sign
up in the box at the top right corner of the online article.
Notes
To request permissions go to:
http://group.bmj.com/group/rights-licensing/permissions
To order reprints go to:
http://journals.bmj.com/cgi/reprintform
To subscribe to BMJ go to:
http://group.bmj.com/subscribe/